With fall and holiday planning well under way, how do mailers get more out of the cooperative database, increase response rates, and find new sources of prospecting names? Here are several advanced tactics:
Response modeling. Most co-op models that have been used in the past are based solely on modeling those who responded. As a result, the models are based only on the predictive characteristics of those individuals who purchased from your offer. When response models are created, both the responders (buyers) and the nonresponders (nonbuyers) are profiled. The models are therefore created using both predictive characteristics for buying behavior as well as characteristics that are predictive for nonbuying behavior. Thus, these models are among the top-performing models in the mail. Abacus and I-Behavior will create response models upon request. The newest consumer cooperative database, Wiland, has based all its modeling on response models.
Dynamic batch optimization. Optimization has allowed mailers to identify inactive house file buyers who would not otherwise be mailed. In many cases, you can take only the top two segments of the model before the results are such that they do not meet the cut-off criteria for the mailing. Regular optimization models are based on RFM (recency/frequency/monetary value) characteristics within the entire co-op database. Dynamic batch optimization, offered by Abacus, is based on category affinity, which is more predictive than RFM of the database. In addition, it can be run in rounds based on synergistic categories for each round. For instance, if you sold hard goods for the home, you may run the first round of optimization based on furniture. After removing the names you wish to mail, rerun the optimization based on home decor. Again remove the names you wish to mail and continue the cycle with lighting, and so on. By running dynamic batch optimization, you will be able to identify and mail more housefile names that will respond at a higher level.
Reactivation modeling. Reactivation modeling, offered by NextAction, is also a way to identify more house file buyers who would not otherwise have been mailed. Unlike typical optimization models that are simply built on RFM within the co-op database of your older buyers, reactivation models are built off response models just like the prospecting models you are using. Predictive characteristics within the active buyer file are used to identify the inactive buyers most likely to respond.
Responder files. Basing the models on appropriate responder (buyer) files is critical to reaching optimal performance. First you need to verify that you are not including Internet-generated buyers in the responder file. These buyers are known to be buyers with lower lifetime value. By including these names in the responder file, the resulting models will be inherently weaker. Second, if you have a diverse product line, you should break out the buyers into product category responder files. For instance, a gardening cataloger might break the responder file into groups such cas structure (greenhouses, sheds, etc.), gardening accessories, and live goods (plants). These models can be tested to determine the strength of each category.
By continually testing new models and new techniques, the co-op names you are renting will perform better and better. Models and modeling techniques are being invented and tested on a regular basis. With most prospecting now originating from the cooperative databases, it is critical to understand the models offered and how to use them to drive the most profitable sales possible.
Michelle Farabaugh is a partner with San Francisco-based Lenser and will be presenting “Maximizing Lifetime Value: The Promise of Multichannel Marketing” at MCM Live, a two-day intensive session presented by MULTICHANNEL MERCHANT, June 1-2 in New York. For more information on MCM Live, which explores the multichannel organization from the inside out and from the front end to the back, go to www.MCMLive.com.